PostgreSQL monitoring that never touches your data
A read-only Rust daemon runs inside YOUR infrastructure, collects system catalog metrics every 30 seconds, and pushes them out. No inbound connections. No access to your actual rows. Ever.
Free for 1 database — no credit card
[Your Infrastructure] [PG Sentinel Cloud] +-----------------------+ +--------------------+ | PostgreSQL DB | | Dashboards | | +------------------+ | metrics | +--------------+ | | | pg_stat_* |-+--push--->| | Queries | | | | system catalogs | | (TLS) | | Indexes | | | +------------------+ | | | Bloat | | | | | | Vacuum | | | +------------------+ | | +--------------+ | | | Rust Daemon | | +--------------------+ | | ~5MB binary | | | | every 30s | | | +------------------+ | +-----------------------+
How it works
Three steps. Five minutes. No config files to write.
Install the daemon
$ docker run -d \
-e PG_HOST=your-db.rds.amazonaws.com \
-e PG_USER=pg_sentinel \
-e PG_DATABASE=myapp \
-e API_TOKEN=your_token \
dlto/pg-sentinel-daemon
Metrics flow every 30 seconds
The daemon reads pg_stat_statements, index stats, table sizes, vacuum progress, and more. All from system catalogs. No access to your application data.
Open your dashboard
Pre-built views for queries, indexes, bloat, connections — data's already flowing. No dashboards to configure, no queries to write.
What you get
Every metric pulled from PostgreSQL system catalogs, organized into actionable dashboards.
Query Performance
See exactly which queries are burning CPU and I/O — pulled straight from pg_stat_statements, without you ever logging in. Track mean and P95 execution times, calls per minute, and rows returned. Sort by total time to find the queries that matter most.
Index Health
Find unused indexes wasting disk, missing indexes slowing queries, and bloated indexes needing a reindex — detected automatically from system catalog snapshots. Get actionable recommendations with the exact SQL to run.
Table Bloat & Vacuum
Know when autovacuum stalls before bloat becomes a problem. Track dead tuples, vacuum progress, and transaction ID wraparound risk. See which tables need attention right now, not after your database slows to a crawl.
Connection Monitoring
See active, idle, and waiting connections broken down by application. Spot connection leaks and lock contention in real time. Know exactly which client is holding the lock that's stalling your deploy.
"We used to spend Friday afternoons SSH-ing into production to check vacuum progress. PG Sentinel replaced that with a 30-second glance at a dashboard."
Jamie Rodriguez
Staff Engineer, Infrastructure
What the daemon can and cannot see
Security is not a feature we bolted on. It's the architecture.
What we read
- pg_stat_statements
- pg_stat_user_tables
- pg_stat_user_indexes
- pg_stat_activity
- System catalogs only
What we never see
- Your actual table data
- Query parameters
- Row contents
- Application secrets
- Credentials or PII
How data flows
Push-only. No inbound connections to your database. Ever.
Frequently asked questions
- No. The daemon only reads PostgreSQL system catalogs like pg_stat_statements and pg_stat_user_tables. It never runs SELECT queries against your application tables. Your row data, query parameters, and application secrets never leave your infrastructure.
- Single binary, Docker container, or package manager. The daemon is a ~5MB Rust binary with no runtime dependencies. Run it as a Docker sidecar, a systemd service, or directly on the command line. Setup takes under five minutes.
- PostgreSQL 12 through 17. The daemon adapts its queries based on the version it detects, so newer catalog columns are used when available without breaking compatibility with older releases.
- No. The daemon makes outbound HTTPS connections only. It pushes metrics to the PG Sentinel API. There are no inbound connections, no VPN tunnels, and no firewall rules to configure.
Does PG Sentinel access my actual data?
How is the daemon installed?
What PostgreSQL versions are supported?
Do I need to open any ports?
Your Postgres is already talking. Start listening.
Free for 1 database — no credit card required